SEARCH

SEARCH BY CITATION

Dear Editor,

We thank Dr. Mariam El-Zein and coworkers for their interest in our article. After reviewing the comments and checking our raw data and original article carefully, we are pleased to reply and explain their questions one by one.

Heterogeneity

In our meta-analysis, significant heterogeneity was indeed observed in almost all the study results, even, stratified analyses by study design, smoking status, area, and other confounders could not eliminate the heterogeneity. As Dr. Mariam El-Zein pointed out, a lot of factors could contribute to the heterogeneity such as distinctive characteristics of individual studies, variability between studies and also a large number of studies, so it is sometimes difficult to find the accurate source of heterogeneity. However, the inverse association kept unchanged with stratification by the confounders. What is more, the pooled RR did not change much (figure not shown) when the sensitivity analysis was conducted. In addition, when a new meta-analysis was carried out after removing the studies that caused significant heterogeneity (refer to the Refs. 7, 9, 10, 13, 16, 18, 26, 28, 36, 41, and 42 in the original article), the results showed that the pooled RR was 0.78 (95%CI: 0.75–0.81, I2= 0.0, p = 0.697), which stayed almost the same with the original one (figure not shown). All of the evidence above indicated that the heterogeneity exerted little influence to our meta-analysis results.

Methodological Quality of Original Studies

We acknowledge that we did not include a formal evaluation table of study quality in our meta-analysis, because it is quite common in the similar meta-analysis.1,2 Although there are many appraisal tools to evaluate the methodological quality of studies; however, for the observational studies, there are still not well-established standard assessment methods like JADAD or JUNI for random controlled trials.3 As all the studies included in our study were in accordance with our study selection criteria, we did not exclude the “smaller” studies, which was a relative concept. It is sometimes incompatible to choose a comprehensive analysis or to get an improved result after excluding some “smaller” studies.

Confounding by Smoking

As pointed by Dr. Mariam El-Zein, the relationship between BMI and lung cancer risk in nonsmokers is always the central issue that controversy focus on. In our meta-analysis, the results showed that the inverse association existed in female nonsmokers, whereas it became statically nonsignificant in male nonsmokers. As smoking status is the most common confounder in the study of the association between BMI and lung cancer risk, we noticed that the sample size was reduced a lot after focusing on nonsmokers in most original studies, in such case the error could not be avoided. It is a good advice to take a more comprehensive approach to represent life time smoking behavior in the future study as suggested by Dr. Mariam El-Zein.

Recently, we noticed that Smith et al.4 also found an inverse association between BMI and lung cancer risk in smokers in a large cohort study, along with Tarnaud's findings,5 the inverse relationship in smokers is certain and confirmable to some extent, so our extrapolated suggestion that “smokers should improve their nutritional status and maintain a suitable body weight” is tenable and apprehensible. However, the relationship between BMI and lung cancer risk in nonsmokers is still ambiguous, especially in males, we speculate that even such relationship exists, the extent is relatively weak, and obviously there is no evidence that obesity can increase lung cancer risk like some other types of cancer in nonsmokers. For future studies, on one hand, strict control for confounding factors especially smoking is important, besides, it is also necessary to conduct well-designed animal experiments to confirm the association and explore the mechanism under it.

In conclusion, the comments from Dr. Mariam El-Zein et al. were pertinent and constructive, which was helpful to interpret the results of the meta-analysis and for the studies relevant in the future.

Yours sincerely,

Yang Yang

Yang Jiao

References

  1. Top of page
  • 1
    He FJ, Nowson CA, MacGregor GA. Fruit and vegetable consumption and stroke: meta-analysis of cohort studies. Lancet 2006; 367: 3206.
  • 2
    Larsson SC, Wolk A. Overweight and obesity and incidence of leukemia: a meta-analysis of cohort studies. Int J Cancer 2008; 122: 141821.
  • 3
    Sanderson S, Tatt ID, Higgins JP. Tools for assessing quality and susceptibility to bias in observational studies in epidemiology: a systematic review and annotated bibliography. Int J Epidemiol 2007; 36: 66676.
  • 4
    Smith L, Brinton LA, Spitz MR, et al. Body mass index and risk of lung cancer among never, former, and current smokers. J Natl Cancer Inst 2012; 104: 77889.
  • 5
    Tarnaud C, Guida F, Papadopoulos A, et al. Body mass index and lung cancer risk: results from the ICARE study, a large, population-based case–control study. Cancer Causes Control 2012; 23: 111326.